Skip to content

NEWS.md

Latest commit

 

History

History
142 lines (124 loc) · 6.59 KB

NEWS.md

File metadata and controls

142 lines (124 loc) · 6.59 KB
Raw

tidygraph (development version)

tidygraph 1.3.1

  • Fix a bug in tbl_graph() when edge to and from where encoded as factors
  • Secure compitability with igraph 2.0.0

tidygraph 1.3.0

  • Add resolution argument to group_louvrain() to mirror the igraph function
  • as_tbl_graph() on an edge dataframe now only adds a name node attribute if the edges are encoded as a character (#147)
  • Added node_is_connected() to test whether a node is connected to a set of nodes (#165)
  • Deprecated play_erdos_renyi() in favour of play_gnm() and play_gnp() (#152)
  • Added the whole family of slice_*() functions from dplyr (#128)
  • Added methods for tidyr::replace_na() and tidyr::drop_na() (#114)
  • Added edge_is_bridge() for querying whether an edge is a bridge edge (#113)
  • Added a glimpse() method for tbl_graph and morphed_tbl_graph objects (#30)
  • Add iterate_n() and iterate_while() to perform repeated modifications of a graph for a specific number of times or until a condition no longer is met (#43)
  • Add focus()/unfocus() verbs to limit node and edge algorithms to a subset while still keeping the full graph context (#18)
  • Data frame subclasses with sticky columns (such as those from sf and tsibble) now works with the tbl_graph constructors (#184)
  • graph_automorphisms() gains a color argument in line with capabilities in igraph
  • graph_mean_dist() now supports edge weights through a new weights argument
  • Added to_largest_component() morpher
  • Added graph_is_eulerian() and edge_rank_eulerian() for eulerian path calculations
  • Added to_random_spanning_tree() morpher
  • Added min_order argument to to_components() morpher
  • Added random_walk_rank() to perform random walks on the graph
  • Added centrality_harmonic() + deprecated centrality_closeness_harmonic(). The latter is an interface to netrankr while the former is a more efficient and flexible igraph implementation.
  • Added group_color() as an interface to greedy_vertex_coloring() in igraph
  • Added group_leiden() to interface with cluster_leiden() in igraph
  • Added group_fluid() to interface with cluster_fluid_communities() in igraph
  • Added edge_is_feedback_arc() to interface with feedback_arc_set() in igraph
  • Added graph_efficiency() and node_effeciency() interfacing with global_efficiency() and local_efficiency() in igraph

tidygraph 1.2.3

  • Small updates to work with new versions of igraph and dplyr

tidygraph 1.2.2

  • Activating a grouped tbl_graph with what is already active will no longer cause grouping to be dropped (#121)
  • General upkeep

tidygraph 1.2.1

  • Move compiled code to cpp11
  • Improve messaging with rlang and cli
  • New feature: the following hierarchical clustering functions group_edge_betweenness, group_fast_greedy, group_leading_eigen and group_walktrap have a new argument n_groups that controls the numbers of groups computed. The argument expects an integer value and it is NULL by default.

tidygraph 1.2.0

  • graph description now recognise undirected trees
  • Added pkgdown site at https://tidygraph.data-imaginist.com
  • Prepare tidygraph for dplyr 1.0.0 (#118 and #119)
  • Add possibility of controlling which column in nodes are used for matching if the to and from columns in edges are character vectors during construction (#89)
  • bind_graph() now accepts a list of graphs as its first argument (#88)
  • Add graph_modularity() for calculating modularity contingent on a node grouping (#97)
  • Edge weights are now handled more consistently to avoid igraph using a possible weight edge attribute. weights = NULL will always mean that no edge weight is used (#106).
  • Neighborhood graph in map_local() and siblings will now contain a .central_node node attribute that will identify the node from which the local graph has been calculated (#107)

tidygraph 1.1.2

  • Compatibility with dplyr 0.8

tidygraph 1.1.1

  • Better conversion of network objects. Old conversion could mess up edge attributes.
  • Changes to anticipate new version of tibble and dplyr
  • tibble-like dimming of non-data text in printing
  • Edge-length is now preserved when converting from phylo
  • Added to_subcomponent morpher to work with a single component containing a specified node
  • Morphers that reference nodes now correctly tidy eval the node argument
  • Add node_is_adjacent to query which nodes are directly connected to a set of nodes
  • Add fortify method for tbl_graph object for plotting as regular data with ggplot2

tidygraph 1.1.0

  • Fix bug when coercing to tbl_graph from an adjacency list containing NULL or NA elements.
  • Change license to MIT
  • Add convert verb to perform both morph and crystallise in one go, returning a single tbl_graph
  • When collapsing edges or nodes during morph the original data will be stored in .orig_data instead of .data to avoid conflicts with .data argument in many tidyverse verbs (BREAKING)
  • as_tbl_graph.data.frame now recognises set tables (each column gives eachs rows membership to that set)
  • Add with_graph to allow computation of algorithms outside of verbs
  • graph_is_* set of querying functions has been added that all returns logical scalars.
  • Add %N>% and %E>% for activating nodes and edges respectively as part of the piping.
  • mutate now lets you reference created columns in graph algorithms so it behaves in line with expected mutate behaviour. This has led to a slight performance decrease (millisecond scale). The old behaviour can be accessed using mutate_as_tbl where the graph will only get updated in the end.
  • When using to_subgraph with edges, isolated nodes are no longer deleted
  • bind_graphs now work with a single tbl_graph
  • Added .register_graph_context to allow the use of tidygraph algorithms in external functions.
  • Added to_unfolded_tree, to_directed, and to_undirected morphers
  • Add the node_rank_* family of algorithms for seriation of nodes
  • Added to_hierarchical_clusters morpher to work with hierarchical representations of community detection algorithms.
  • All group_* algorithms now ensure that the groups are enumerated in descending order based on size, i.e. members of the largest group/community will always have 1, etc.
  • Fix a bug when filtering all nodes or edges where no nodes/edges would be removed (#42)
  • Added interface to netrankr resulting in 19 new centrality scores and a manual mode for composing new centrality scores
  • Added edge_is_[from|to|between|incident]() to help find edges related to certain nodes